I am a Vero-board type myself because it leaves me the flexibility I need. The LM317 kits I have seen use small insufficient heatsinks. You need good heatsinks, eventually a metallic casing you can use as heatsink (be aware of the needs to insulate the chip cooling pads). You may also need some experience with Vero-board.
I am a Vero-board type myself because it leaves me the flexibility I need. The LM317 kits I have seen use small insufficient heatsinks. You need good heatsinks, eventually a metallic casing you can use as heatsink (be aware of the needs to insulate the chip cooling pads). You may also need some experience with Vero-board.
no experience with vero board but this is just a new challenge, i think is going to be fun anyway. Now i have to start looking for the parts and make a list of the stuff that i really need. BRB
no experience with vero board but this is just a new challenge, i think is going to be fun anyway. Now i have to start looking for the parts and make a list of the stuff that i really need. BRB
If you decide for Vero-board, use single sided fiber glass types.
Pertinax (brown Bakelite) boards quickly loose the PCB pads when heated and double-sided fiber-glass becomes a nuisance because the solder flows through the unused via's and create solder-bubbles on the component side.
well at this stage i don't really have better options...or i do...?! yep i will pay attention, thank you
Mrdrewk,
In case you're still interested, check out these modules: Audio Selbstbau, Thel, Endstufen, Vorstufen, Aktivweichen, analoge Netzteile, Netzfilter.
You can also get some of the same modules in customizable (cheaper) kits: Bausaetze.
Haven't tried these yet but will do so in a few weeks to replace a couple of wall warts. I'm a Yank living in the Frankfurt area since 2014 and THEL is located a few miles away from my little village. I'll go there to pick up the units.
Audiophonics also has some interesting and economical modules: AUDIOPHONICS PSU S3-LP Linear Power supply Module DC LT1083 - Audiophonics.
Since linear PSUs seem to be the latest fad for us audio nuts I've been investigating and have come away quite shocked at what people are paying for this stuff, especially for PC supplies. Smelling the blood in the water, the Chinese have entered the market big time, with "big" being the operative word on prices. Hopefully, a reasonable diy solution isn't far off.
In case you're still interested, check out these modules: Audio Selbstbau, Thel, Endstufen, Vorstufen, Aktivweichen, analoge Netzteile, Netzfilter.
You can also get some of the same modules in customizable (cheaper) kits: Bausaetze.
Haven't tried these yet but will do so in a few weeks to replace a couple of wall warts. I'm a Yank living in the Frankfurt area since 2014 and THEL is located a few miles away from my little village. I'll go there to pick up the units.
Audiophonics also has some interesting and economical modules: AUDIOPHONICS PSU S3-LP Linear Power supply Module DC LT1083 - Audiophonics.
Since linear PSUs seem to be the latest fad for us audio nuts I've been investigating and have come away quite shocked at what people are paying for this stuff, especially for PC supplies. Smelling the blood in the water, the Chinese have entered the market big time, with "big" being the operative word on prices. Hopefully, a reasonable diy solution isn't far off.
Follow up on post #6. It is not that hard to build a linear PSU. It is hard(er) to build a safe and excellent measuring linear PSU. Always use metal casings with a filtered IEC inlet and a real power switch (officially both poles should be switched). Casing should be connected to PE, the DC parts GND can be connected with a 10 Ohm 1 W resistor to PE. Always use M3 bolt/nut and washers for the PE connection to the metal casing and always verify things make good contact. Always use fuse holders and choose the right value fuse. Easiest is a filtered IEC inlet with built in fuse holder and power switch so you have less metalwork to do.
Make sure you use low noise regulators or a discrete circuit that is low noise. Avoid old generic regulators that are noisy. The filter capacitors should be chosen so that ripple voltage is low at the highest load. Rule of thumb is 2200 µF/A for 1V ripple. When possible use CLC filtering.
You will find out that modern low noise regulators are often low drop out types which is even better. Discrete circuits can be better but IC regulators circuits are simple and can be just as good or even better. If you carefully choose transformer voltage you can adapt things to each other and have least heat generation and good quality regulation. Best transformers are R-Core types followed by EI types and last toroids. The latter are cheapest and certainly not the best. Choose the transformer to be 2x the required load ratings.
The combination of things can be a true improvement on generic SMPS that are delivered with equipment but the devil is in the details as always. With todays RF signals in the air shielding and filtering have become necessities. And please note that using an LPS with device A does not mean the SMPS of device B is not polluting mains in your home....
Make sure you use low noise regulators or a discrete circuit that is low noise. Avoid old generic regulators that are noisy. The filter capacitors should be chosen so that ripple voltage is low at the highest load. Rule of thumb is 2200 µF/A for 1V ripple. When possible use CLC filtering.
You will find out that modern low noise regulators are often low drop out types which is even better. Discrete circuits can be better but IC regulators circuits are simple and can be just as good or even better. If you carefully choose transformer voltage you can adapt things to each other and have least heat generation and good quality regulation. Best transformers are R-Core types followed by EI types and last toroids. The latter are cheapest and certainly not the best. Choose the transformer to be 2x the required load ratings.
The combination of things can be a true improvement on generic SMPS that are delivered with equipment but the devil is in the details as always. With todays RF signals in the air shielding and filtering have become necessities. And please note that using an LPS with device A does not mean the SMPS of device B is not polluting mains in your home....
Last edited:
jean-paul,
Thanks for the advice on the R-Cores. Didn't know that. So, where do I find top quality R-Cores? So far, my Google search has turned up relatively inexpensive stuff on ebay. Until now I was considering the Supreme Audio Grade V2 toroids from Toroidy in Poland, which cost about $75 US for 20VA.
So, am I just paying for the silver can bling, and one of the $20 jobs on ebay is the better way to go?
Confused.
Thanks for the advice on the R-Cores. Didn't know that. So, where do I find top quality R-Cores? So far, my Google search has turned up relatively inexpensive stuff on ebay. Until now I was considering the Supreme Audio Grade V2 toroids from Toroidy in Poland, which cost about $75 US for 20VA.
So, am I just paying for the silver can bling, and one of the $20 jobs on ebay is the better way to go?
Confused.
Because of a combination of luck and coincidence I got hold of a lot of R-core transformers years ago otherwise I would probably never have tried them. Since then I always use them whenever I get the chance.
Many toroids leak and they are too high bandwidth so they don't filter HF/RF as much as I like them too. The fact that many of them can be used for output transformer says something. Most are also very susceptible to DC component on mains voltage which causes annoying mechanical hum. All anecdotical in the eyes of toroid lovers but I experienced it too much to have them at the top of my list. EI types in general are better but then also the challenge to find the good and silent ones. I wish I would not say this but older EU made stuff can be very good.
True is that toroids are easy to produce, easy to mount and that you can find them in any catalog/website. The explains why they have become standard just before SMPS (which will be/is standard also for other than quality reasons). As there is not much choice I also use toroids now and then and the Sedlbauer are good.
I am in audio for a very long time and know how to build a good linear PSU but wanted to get to know the most recent regulators so when I was in quarantine I made around 12 linear PSU's based on various regulators and all perform better than the SMPS they are ought to replace. This with regards to noise, RF/mains pollution, HF/RF susceptivity and subjectively (so soundwise). Point is that they are more expensive and it requires careful design to have them performing excellent. My own forgotten SBT PSU as published here still holds up quite OK in the 3A range.
Many toroids leak and they are too high bandwidth so they don't filter HF/RF as much as I like them too. The fact that many of them can be used for output transformer says something. Most are also very susceptible to DC component on mains voltage which causes annoying mechanical hum. All anecdotical in the eyes of toroid lovers but I experienced it too much to have them at the top of my list. EI types in general are better but then also the challenge to find the good and silent ones. I wish I would not say this but older EU made stuff can be very good.
True is that toroids are easy to produce, easy to mount and that you can find them in any catalog/website. The explains why they have become standard just before SMPS (which will be/is standard also for other than quality reasons). As there is not much choice I also use toroids now and then and the Sedlbauer are good.
I am in audio for a very long time and know how to build a good linear PSU but wanted to get to know the most recent regulators so when I was in quarantine I made around 12 linear PSU's based on various regulators and all perform better than the SMPS they are ought to replace. This with regards to noise, RF/mains pollution, HF/RF susceptivity and subjectively (so soundwise). Point is that they are more expensive and it requires careful design to have them performing excellent. My own forgotten SBT PSU as published here still holds up quite OK in the 3A range.
Last edited:
I will add to the clever words of Jean-Paul that an issue to address up-front is what are the most important qualities you want the linear power supply to have. You do not get all qualities in the same design.
If you design for use with a high quality DAC or microphone amplifier that may be supply-rail-noise sensitive, you can use a shunt-regulator design. You do not use a shunt-regulator design if the load has a high and varying consumption. The advantage of the shunt regulator is that it can be designed with low noise and good step response. You may consider the new generation of low-noise regulators though they are expensive and less noise than what is needed is just more expenses than required without a further advantage.
For a load needing a rather constant voltage but having a heavily varying consumption, like a class AB power amplifier, a more simple design with good dynamic properties (load response) is suited. Particular low-noise designs are useless among other because the dynamic consumption will cause much higher voltage swings than the regulator noise. A design with an emitter output I find is suited because of the intrinsically low output impedance of an emitter, where the regulation loop only needs to do little correction when the load current changes importantly.
With an input voltage not much higher (or lower for negative) than the output voltage, a low-drop design, typically with a collector output, may be a necessity. Such a situation is 3.3V out from a 5V input. Reduced heating than for the other configurations but typically with less good load-response compared to an emitter output and more noise than for the shunt-regulator. You don't get all in one.
SMPS designs equally have some very good properties, like very good efficiency and power handling levels, but also some drawbacks like noise and sometimes step-response.
So, for a start it is very important to identify your supply needs and following you choose a regulator concept that is suited. Let's eventually discuss your needs for a particular application here in this thread. I have noticed a tendency among certain members to ask for the “one-fits-all” best regulator design, typically assuming that low-noise is most important. In most cases, low-noise or “standard” doesn't make a difference except for the price. Regulator design is, as written by Jean-Paul, not terribly difficult if you understand a few basic principles. One important basic principle is that you need to know your regulator requirements before choosing the regulator concept. Just like when you buy a car that either needs to be practical, or comfortable, or economical, or easy to park in the city or sporty and no model fits all uses.
Describe your use and we can advice you a suitable regulator concept.
If you design for use with a high quality DAC or microphone amplifier that may be supply-rail-noise sensitive, you can use a shunt-regulator design. You do not use a shunt-regulator design if the load has a high and varying consumption. The advantage of the shunt regulator is that it can be designed with low noise and good step response. You may consider the new generation of low-noise regulators though they are expensive and less noise than what is needed is just more expenses than required without a further advantage.
For a load needing a rather constant voltage but having a heavily varying consumption, like a class AB power amplifier, a more simple design with good dynamic properties (load response) is suited. Particular low-noise designs are useless among other because the dynamic consumption will cause much higher voltage swings than the regulator noise. A design with an emitter output I find is suited because of the intrinsically low output impedance of an emitter, where the regulation loop only needs to do little correction when the load current changes importantly.
With an input voltage not much higher (or lower for negative) than the output voltage, a low-drop design, typically with a collector output, may be a necessity. Such a situation is 3.3V out from a 5V input. Reduced heating than for the other configurations but typically with less good load-response compared to an emitter output and more noise than for the shunt-regulator. You don't get all in one.
SMPS designs equally have some very good properties, like very good efficiency and power handling levels, but also some drawbacks like noise and sometimes step-response.
So, for a start it is very important to identify your supply needs and following you choose a regulator concept that is suited. Let's eventually discuss your needs for a particular application here in this thread. I have noticed a tendency among certain members to ask for the “one-fits-all” best regulator design, typically assuming that low-noise is most important. In most cases, low-noise or “standard” doesn't make a difference except for the price. Regulator design is, as written by Jean-Paul, not terribly difficult if you understand a few basic principles. One important basic principle is that you need to know your regulator requirements before choosing the regulator concept. Just like when you buy a car that either needs to be practical, or comfortable, or economical, or easy to park in the city or sporty and no model fits all uses.
Describe your use and we can advice you a suitable regulator concept.
Noise unimportant; low voltage drop unimportant; drift and other precision characteristics unimportant; dynamic loading to be averaged by decoupling capacitors. In my view, a simple buffered LM317 or even a buffered 7812 with a diode in the reference terminal will do the job.
In such a case, the servers will perform no better or worse than with the wall-warts. A typical supply for such use is an SMPS, eventually of better quality than (some) wall-warts for reliability reasons.
Here, racing is only performed on the national and district roads - unfortunately with serious accidents from time-to-time.
In such a case, the servers will perform no better or worse than with the wall-warts. A typical supply for such use is an SMPS, eventually of better quality than (some) wall-warts for reliability reasons.
Here, racing is only performed on the national and district roads - unfortunately with serious accidents from time-to-time.
Last edited:
When choosing a linear PSU: noise moderate important (is way less than original SMPS anyway), low drop important (otherwise unnecessary and more expensive large heatsink/larger thus more expensive casing/useless heat), low output impedance very important, drift and other precision characteristics unimportant.
Both LM317 and 7812 are not adequate for 12V/3.5A buffered or not. It will result in even more heat generation. LT1084 would be adequate and affordable.
* When the servers are only serving data and no DAC is connected SMPS are likely to be preferred. Things change when a DAC is connected or when digital or (don't do this) analog outputs are used.
Both LM317 and 7812 are not adequate for 12V/3.5A buffered or not. It will result in even more heat generation. LT1084 would be adequate and affordable.
* When the servers are only serving data and no DAC is connected SMPS are likely to be preferred. Things change when a DAC is connected or when digital or (don't do this) analog outputs are used.
Last edited:
Intriguing, and money saving, considering all the DC straightener-outer devices that are marketed by companies like iFI...for prices way higher than the Euro 2.88 LM317 I just found on the Pollin site! Thanks, I will definitely give it try.
Racing on public roads has always been a big problem back in the states. Not here in Germany. People easily get their speed rocks off on the Autobahn. No speed limit, but stunts not allowed. Precision. Alertness. Very few accidents.
Racing on public roads has always been a big problem back in the states. Not here in Germany. People easily get their speed rocks off on the Autobahn. No speed limit, but stunts not allowed. Precision. Alertness. Very few accidents.
LM317 is 1.5A and at least 30 years old thus high dropout which will mean useless heat generation/power dissipation. Maybe it helps to discuss 1 subject at a time
Companies that manufacture SMPS know their products limitations and audiophile SMPS manufacturers also know the audiophiles limitations.
Companies that manufacture SMPS know their products limitations and audiophile SMPS manufacturers also know the audiophiles limitations.
Last edited:
I will claim that I can manage to get 10A regulated out from an LM317 based regulator when using a TIP3055 buffer transistor in a TO-247 housing (my first drawing below).
The TIP3055 buffer transistor will handle almost all the load current and take almost all the drop-power. The LM317 only supplies a stable base-voltage to the TIP3055 and in the order of 1/40-ties of the load current plus the bias current set by the base-emitter resistor of the TIP3055. When the TIP3055 sees a large change in the load-current, the LM317 output sees a much smaller change in TIP3055 base-current and with the fixed bias current (base-emitter resistor), a relatively smaller change. Therefore the output voltage change will be rather small even for a huge change in load current because the output voltage drop is determined by the TIP3055 base-emitter voltage change. The change will hardly exceed a couple of hundreds of millivolt.
Yes, with a low-drop regulator we can perhaps operate with less voltage drop than with an LM317 but the difference is not as important as we could expect. The reason is that the largest change in voltage to be accounted for by the minimum voltage drop of the regulator is the drop in voltage from the rectified transformer - from almost unloaded to full load (my second drawing below). For a 12V supply, this input voltage variation is likely to be in the order of 3-4V. The buffered LM317 shown in the first drawing can work down to an input-output voltage drop of some 2.5V. Which low-voltage-drop regulator can supply many Amperes with a drop much below 2.5V?
Linear regulators are inherently suffering from power loss even if a low-drop regulator is used.
We can replace the transformer+rectifier circuit with an SMPS having an output voltage some three Volts above the output voltage of the linear regulator. Then we can reduce the power loss in the linear regulator but the complexity is increased. Next, soon we will question if we really need the downstream linear regulator or if the SMPS can do on its own.
Yes, the LM317 is an old design with more noise than some modern (and more expensive) ICs. 0.003% noise (ST; less than a millivolt on a 12V output) is absolutely no problem for the majority of general purpose uses. Old, but still good enough (my wife says the same about me). Thus, a buffered LM317 can handle surprising current levels with a sufficient “quality” to suit most of our general purpose needs. The total price is very modest.
However, when adding the buffer transistor at the output of the LM317, the LM317 current limiting function is lost. This is why I added a resistor in series with the collector of the buffer transistor (my second drawing below). Using a PNP base-emitter to detect when the resistor voltage gets close to 0.7V leaves us the possibility to pinch-off the LM317 reference terminal voltage and thereby have the current limiting functionality back.
LM317 is not for delicate low power applications but it is very useful for most general purpose applications with higher consumption.
My defense for the LM317.
The TIP3055 buffer transistor will handle almost all the load current and take almost all the drop-power. The LM317 only supplies a stable base-voltage to the TIP3055 and in the order of 1/40-ties of the load current plus the bias current set by the base-emitter resistor of the TIP3055. When the TIP3055 sees a large change in the load-current, the LM317 output sees a much smaller change in TIP3055 base-current and with the fixed bias current (base-emitter resistor), a relatively smaller change. Therefore the output voltage change will be rather small even for a huge change in load current because the output voltage drop is determined by the TIP3055 base-emitter voltage change. The change will hardly exceed a couple of hundreds of millivolt.
Yes, with a low-drop regulator we can perhaps operate with less voltage drop than with an LM317 but the difference is not as important as we could expect. The reason is that the largest change in voltage to be accounted for by the minimum voltage drop of the regulator is the drop in voltage from the rectified transformer - from almost unloaded to full load (my second drawing below). For a 12V supply, this input voltage variation is likely to be in the order of 3-4V. The buffered LM317 shown in the first drawing can work down to an input-output voltage drop of some 2.5V. Which low-voltage-drop regulator can supply many Amperes with a drop much below 2.5V?
Linear regulators are inherently suffering from power loss even if a low-drop regulator is used.
We can replace the transformer+rectifier circuit with an SMPS having an output voltage some three Volts above the output voltage of the linear regulator. Then we can reduce the power loss in the linear regulator but the complexity is increased. Next, soon we will question if we really need the downstream linear regulator or if the SMPS can do on its own.
Yes, the LM317 is an old design with more noise than some modern (and more expensive) ICs. 0.003% noise (ST; less than a millivolt on a 12V output) is absolutely no problem for the majority of general purpose uses. Old, but still good enough (my wife says the same about me). Thus, a buffered LM317 can handle surprising current levels with a sufficient “quality” to suit most of our general purpose needs. The total price is very modest.
However, when adding the buffer transistor at the output of the LM317, the LM317 current limiting function is lost. This is why I added a resistor in series with the collector of the buffer transistor (my second drawing below). Using a PNP base-emitter to detect when the resistor voltage gets close to 0.7V leaves us the possibility to pinch-off the LM317 reference terminal voltage and thereby have the current limiting functionality back.
LM317 is not for delicate low power applications but it is very useful for most general purpose applications with higher consumption.
My defense for the LM317.
Attachments
Dinosaur LM723 with a pass transistor is lower noise in general at least in my projects. It is more elegant, has overcurrent and short circuit protection too which is simply good design practice. "Good enough" is a strange way of looking at matters when there is nothing against lower noise when it does not cost more etc. On the contrary I would say. If "good enough" exists then "not good enough" and "better than required" exist as well. If I would have to choose and stuff costs about the same it would be an easy choice
The low drop relatively low noise LT1084 (5A !) can be found for 2 Euro a piece and adds simplicity to the project. It has enough protection as well.
Both devices have the upper hand IMHO. I know LM317 is kind of a religion and should be used in every device thinkable and the drawers of DIYers are apparently full with them. However, there are other regulators out there when higher than 1.5A currents are needed.
The low drop relatively low noise LT1084 (5A !) can be found for 2 Euro a piece and adds simplicity to the project. It has enough protection as well.
Both devices have the upper hand IMHO. I know LM317 is kind of a religion and should be used in every device thinkable and the drawers of DIYers are apparently full with them. However, there are other regulators out there when higher than 1.5A currents are needed.
Last edited:
In my opinion, toroidal transformers are unfairly maligned here and certainly not the worst choice for most audio equipment. I will gladly deal with the higher interwinding capacitance in exchange for lower stray fields.
Opinions and beliefs are OK (in politics and church) but please compare R-Core with toroid for factual information. Toroids are not maligned but they are objectively (from electrical point of view) not the most optimal choice. I wrote "cheapest and not the best" regarding toroids which is fairly honest I think. See post #29 for details. In practice one can choose between toroid and R-core transformers as EI types in audio VA ratings are not easy to find anymore in excellent quality.
If one does not know anything else toroids are fine and they can be found at every supplier. What is this with "good enough" and defending stuff? Why would done limit ones choices when one wants best results?
You can mount a circuit almost against an R-core transformer without having issues with stray fields. I handled and tested a very large number of them and never experienced one that hums mechanically (this is not the case with many toroids and a small DC component on the mains voltage). These are good properties in audio.
If one does not know anything else toroids are fine and they can be found at every supplier. What is this with "good enough" and defending stuff? Why would done limit ones choices when one wants best results?
You can mount a circuit almost against an R-core transformer without having issues with stray fields. I handled and tested a very large number of them and never experienced one that hums mechanically (this is not the case with many toroids and a small DC component on the mains voltage). These are good properties in audio.
Last edited:
Objectively, toroids will have less stray magnetic field than even an R Core. Proper power supply design mitigates the effects of the increased interwinding capacitance. One of these problems is much easier to solve than the other. I have designs at work with high performance ADCs that would require more distance from the transformer or attempts at magnetic shielding to achieve the same spotless FFT with a split bobbin or R-Core. Have you done any comparison measurements?
R-Core are very good and may be the best choice for some boxes, but there are certainly applications where I'd rather use a toroid or even SMPS.
R-Core are very good and may be the best choice for some boxes, but there are certainly applications where I'd rather use a toroid or even SMPS.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.